Evidential Occupancy Grid Map (EOGM)

C++ · ROS · 2024

Robotic occupancy mapping from onboard sensors, using Dempster-Shafer evidence theory.

Overview

Based on our professor’s research paper on pratical application of the Dempster-Shafer theory in robotics, we developed a real-time algorithm that creates maps of free and occupied spaces in a robot’s environment using onboard LiDAR, IMU and GPS sensors. It was developed for Robot Operating System (ROS). It’s fast and efficient due to various memory optimizations, vectorization and parallelization techniques.

Engineering approach

Sensor-derived occupancy maps are fused by a ROS node using evidential reasoning. The repository documents AVX2 optimizations, illustrating trade-offs between scientific computing, real-time robotics and low-level performance.

Scope and limitations

Academic or research work, potentially collaborative. It illustrates an engineering method or experiment rather than claiming a commercially deployed product.

Explore the project

Source repository, documentation or official result ↗

Related work: Model railway network · Connected speaker